Plenty of artificial intelligence (AI) agents flop for reasons that have nothing to do with their capabilities. They get sold as human, and people can tell. James Finister’s example is Vodafone. It built an agent, dressed it up to seem human, and watched customers avoid it. Relabeled honestly as an AI, with a name that dropped the pretense, the same agent took off because people knew what they were dealing with.
For James, it’s a small version of a much bigger problem: service management still doesn’t value people enough. It undervalues the humans who do the work, rushes to trust software that imitates them, and treats supplier relationships as contracts.
James is the eleventh guest in Conversations with Giants, Roman Jouravlev’s interview series. After a long career in service management, he now works on AI ethics and the standards meant to govern it.
Put AI Governance in Early, Not After It Goes Wrong
His advice starts with timing. Put AI governance in when you start putting in the AI, not once it’s already going wrong by stealth. Make it actionable, because there are thousands of standards that say what to care about and almost none that say how to do it. Bring a wide range of people into how AI gets used, and stay realistic about what it can and can’t do. The thing he warns about most is anthropomorphism. People take life advice from AI and form relationships with it, and it’s still a computer.
He isn’t commenting from the sidelines. James helps write the international standards for this, including IEEE 7014, the standard on emulated empathy in AI systems, and its follow-on for the empathy people project onto AI agents and companions. He and Roman are also turning a white paper on AI governance into a book written to sit alongside ITIL (Version 5). He wants it usable above all, tailored to an organization’s own risks rather than a generic top ten, and readable, because things that aren’t readable don’t get read.
People Are Not the Problem to Design Out
All of that AI governance is really about protecting people, and that’s where James has the strongest views. He thinks service management has a habit of treating them as the problem to be fixed, and he did it himself for years. On service integration and Agile projects, he was forever asking people to change how they worked, from the standard consulting assumption that when something isn’t working, the people are the problem. At some point, that flipped. You should be designing systems around people, not the other way around. You hired someone for who they are, so turning around and treating that as a fault is, in his words, unethical.
The interest is partly personal. James has neurodiverse traits, and he’s wary of the language that calls neurodiversity a superpower. Still, he’s clear that it brings strengths worth using. As Faith Thomas, Lead IT Service Management Practitioner for the University of Birmingham, argued at a recent ITSM event, if you make things better for neurodivergent people, you generally make them better for everyone. He separates mental health from wellbeing. Wellbeing is the cup of tea and the nice day out. Mental health is a serious clinical matter and a job for specialists, and after losing a family member to suicide, he pushed to get it talked about openly at industry events, where it had been silent for years.
SIAM Is About Relationships, i.e. People, Not Just Contracts
Relationships are the part James thinks people miss about the discipline he spent the longest in. Most organizations can no longer run everything from one internal IT department, which is why SIAM exists to bring one view, one language, and shared risk to a supply chain full of different providers. James spent thirteen years at TCS doing exactly this on global deals, and his conclusion is that SIAM is mostly about relationships and trust. If the people across that supply chain can’t work together, no contract saves it. The discipline has slipped off conference agendas, partly because AI took the spotlight, what Paul Wilkinson calls the new shiny thing that really helps.
Organizations rarely realize how tangled their supply chains are. James has walked into companies that think they have one outsourcing partner and find fifty separate IT contracts, run by different people, with no join-up. Something breaks, they blame the vendor or the tool, they run a three-year replacement, and it breaks again. Often, it was never the vendor and never the tool. It was the model being used to manage the services. And there’s a hard message in that for the internal IT team, because it doesn’t matter that you outsourced the work. It’s still your fault when it goes wrong, because you let the contract and you carried the risk.
The Ethics Won’t Change, But Our Attitude to Them Will
The question of who bears the risk is also what makes James wary of AI. He doesn’t expect AI ethics to shift much. He thinks AI is morally neutral, and the ethics we need for it are largely the ones we already apply to ourselves. What he expects to change is how seriously people take them, driven by what he calls the tombstone imperative – harm is bad enough that ethics stops being a nice-to-have for nice companies. AI is already being used in HR without the governance to match, and it hits real people. The standards themselves are largely shaped by wealthy Western democracies, so many voices go unheard. And the largest AI providers are starting to behave less like tool vendors and more like defense contractors, which is a different game entirely.
The people entering this industry are doing so in hard times, and James thinks the least it owes them is a workplace that treats them as people rather than a problem to be fixed.
Key Takeaways
- Put AI governance in place when you start deploying AI, not after it drifts and goes wrong by stealth. Keep it actionable and tailored to your own risks rather than a generic checklist.
- Be honest that an AI is an AI. Give agents non-human names and clear labels, because people engage more readily when they know what they’re dealing with.
- Design your ways of working around the people you have, rather than treating them as the problem when they don’t fit the system.
- Map your actual supply chain before you blame a vendor or replace a tool. The fault often lies in the operating model, and outsourcing the work never outsources the accountability.
- If you’re early in your career, learn the business you support and speak its language, keep building new skills, and if your employer won’t fund training, fund it yourself.
The full conversation with James Finister is available as episode 11 of Conversations with Giants, hosted by Roman Jouravlev:
Sophie Danby
Sophie is a freelance ITSM marketing consultant, helping ITSM solution vendors to develop and implement effective marketing strategies.
She covers both traditional areas of marketing (such as advertising, trade shows, and events) and digital marketing (such as video, social media, and email marketing). She is also a trained editor.
